Design of the randomized seeding experiment of the WWMPP

The main purposes of the Wyoming Weather Modification Five-Year Pilot Project (WWMPP) are to establish a winter orographic cloud seeding program in three target areas (the Medicine Bow Range, the Sierra Madre Range and the Wind River Range) and to evaluate the feasibility and effectiveness of the seeding. The WWMPP is unique among state-sponsored programs in that it includes a substantial evaluation component.

A large portion of the evaluation effort is the design of a randomized seeding experiment to detect and quantify potential seeding effects. The statistical evaluation of the randomized seeding experiment demands that analyses and procedures be specified in advance (a priori). This paper describes some details of the design of the randomized seeding experiment, including justification for the various aspects and criteria. A companion paper describes some of the statistical issues in more detail.

Some of the main points of the design that will be discussed are:

1. Target areas in the two southern ranges, the Medicine Bows and the Sierra Madres, that are correlated.

2. The placement and coverage of ground-based seeding generator sites.

3. The ability to implement a cross-over design, where one target area is randomly determined to be seeded while the other becomes the control.

4. The length of treatment periods (4-hr periods) and the use of buffer periods (2-hr) to clear the areas of seeding material.

5. The use of high-resolution precipitation measurements in each target area, and control gauges associated with each target.

6. Criteria for case selection, and data required to fulfill these criteria as objectively as possible. It will be necessary to meet these criteria in both targets simultaneously.

7. The use of ratio statistics - described more completely in the companion paper.

8. Estimating the number of paired cases expected per season, and determining an estimate of the number of samples needed for statistical significance (also covered in the companion paper).

9. How to assess issues of targeting uncertainty and contamination potential. This is particularly relevant for the Medicine Bow and Sierra Madre ranges since one is largely upwind of the other.